PETROLEUM SYSTEMS IN THE RIONI AND KURA BASINS OF GEORGIA

The Neogene Rioni and Kura foreland basins in Georgia are located between the converging Greater and Lesser Caucasus fold‐and‐thrust belts. The Rioni Basin continues westward into the Black Sea whereas the Kura Basin extends eastward into Azerbaijan and the Caspian Sea. “Pre‐” and “post‐salt” petroleum systems are distinguished in the Rioni Basin separated by an Upper Jurassic evaporite succession of regional extent. The pre‐salt petroleum system in the northern Rioni Basin is still poorly understood. Bathonian shales have generated oil which has been recorded in Middle Jurassic sandstones. However, as the origin of the oil in Upper Jurassic sandstones (e.g. at the Okumi oil discovery) is still problematic, the pre‐salt petroleum system remains poorly constrained. Gas‐rich, high volatile bituminous coals of Bathonian age may represent a CBM play.


INTRODUCTION
Georgia is located in the South Caucasus between the fold-and-thrust belts of the Greater Caucasus in the north and the Lesser Caucasus in the south (Fig. 1). Two Cenozoic foreland basins separate these mountain ranges: the Rioni Basin to the west, which continues into the Black Sea; and to the east, the Kura Basin which extends toward Azerbaijan and the Caspian Sea (Adamia et al., 2011). The fill of both basins is highly deformed along the fronts of the adjacent fold-andthrust belts (Fig. 2).
Oil and gas seeps in the Rioni and Kura Basins have been known since ancient times, and oil from seeps has been used in Georgia at least since the Medieval period (Nibladze and Janiashvili, 2014). Modern hydrocarbon exploration began in 1869, when the Prussian entrepreneur Walter Siemens obtained licenses for a number of areas in SE Georgia and produced oil from the Mirzaani field (see Fig. 1b for field location; Nibladze and Janiashvili, 2014). However, only a small number of commercial oil fields were subsequently discovered in the period between 1939 and 1974, and no new discoveries have been made since then. Current production is low (<1000 barrels of oil per day; Nachtmann et al., 2015a). This suggests that the petroleum prospects of Georgia are inferior to those of Azerbaijan to the east and those of the Russian (northern) side of the Greater Caucasus, although the regional geology of Georgia in many respects resembles that of those areas. In Georgia, reservoir units with varying but often poor quality are considered to be the main problem for hydrocarbon accumulation and production (Nachtmann et al., 2015a). In addition, recent studies that focused on the principal Eocene and Oligocene source rocks (Pupp et al., 2018;Sachsenhofer et al., 2018a, b;Vincent and Kaye, 2018)

Achara-Trialeti
Caspian Sea K u r a B a s in Baku G re a te r C a u c a s u s L e s s e r C a u c a s u s A n a t o li a n -I r a n ia n P la t e a u   also varies strongly both vertically and between the Rioni and Kura Basins. Apart from conventional oil and gas, Jurassic highvolatile bituminous coal in the Tkibuli-Shaori region (located NW of the Dziruli Massif: Fig. 1b) may have coal-bed-methane (CBM) potential (GIG, 2016). Thermogenic gas and oil stains have been observed in the vicinity of the coal seams (Aslanikashvili and Iorashvili, 2013).
The purpose of the present paper is to present an improved understanding of the petroleum systems in Georgia. To this end, published information relevant to the petroleum geology of the country (e.g. Patton, 1993;Robinson et al., 1997;Nachtmann et al., 2015a;Boote et al., 2018;Yükler et al., 2019) is summarized in the first part of the paper. In the second part, new data is presented for the insufficiently studied Mesozoic and Cenozoic source rock units based on the analysis of a total of 470 rock samples together with analytical results for 19 oil and five gas samples. Most of these new source rock and hydrocarbon samples were made available by Georgian Oil and Gas within the context of their ongoing exploration activities. In a final section, the new data together with published source rock (Pupp et al., 2018) and oil data Mayer et al., 2018) are used to determine oil families, to support oil-source correlations, and to define petroleum systems.

GEOLOGICAL SETTING
Georgia lies within the Alpine-Caucasus-Himalaya orogenic belt, a zone of ongoing collision between the Eurasian and African-Arabian plates. The Caucasus region is divided from north to south into the doublyvergent Greater Caucasus fold-and-thrust belt; the Dziruli Massif, which separates the Rioni Basin in the west from the Kura Basin in the east; and the strongly deformed Lesser Caucasus (Fig. 1a). The Rioni and Kura Basins are Cenozoic foreland basins deformed by the south-verging thrust structures of the Greater Caucasus and by the north-verging structures of the Achara-Trialeti thrust-fold belt (e.g. Banks et al., 1997;Nemcok et al., 2013;Alania et al., 2017;Fig. 2). The geodynamic evolution of the Caucasus area is summarized in Fig. 3.
During Mesozoic time the area of the future Lesser Caucasus formed an island arc (Fig. 3) on the southern margin of the European (Scythian) platform. Apart from volcanic rocks, the Lesser Caucasus comprises Tethyan oceanic rocks as well as Mesozoic to Lower Eocene shallow-water deposits from the carbonatedominated shelf along the Tethyan margin (e.g. Adamia et al., 2015). During the Early Jurassic, sedimentary rocks with some minor volcanic contributions were deposited north of the Lesser Caucasus volcanic arc (e.g. Adamia et al., 2011). Back-arc extension during the late Early Jurassic resulted in significant thinning of the continental lithosphere and increased subsidence in the Great Caucasus Basin (McCann et al., 2010) due to a shallowing of the northward subducting slab (McCann et al., 2010;Fig. 3). Also significant are Bathonian (Middle Jurassic) coal-bearing successions together with Kimmeridgian to Tithonian evaporitic rocks (anhydrite, gypsum, halite) which act as major detachment horizons (Banks et al., 1997;Fig. 4).
During Middle Eocene times, the roll-back of the Neotethyan slab induced volcanic activity and the development of the Achara-Trialeti intra-arc rift south of the Dziruli Massif (Adamia et al., 2011;Gusmeo et al., 2021;Fig. 3).
Coeval with the accumulation of thick volcanic and volcaniclastic rocks in the Achara-Trialeti back-arc basin and the southern and eastern parts of the Dziruli Massif, organic-rich marls of the Kuma Formation ( Fig. 4) were deposited to the north of the Greater Caucasus and in the northern part of the Rioni Basin (Beniamovski et al., 2003;Sachsenhofer et al., 2018b;van der Boon et al., 2019). This formation is overlain by Upper Eocene marls (Belaya Glina Formation) deposited in oxygenated subtropical marine conditions.
Middle Eocene shales interfingering with submarine mass transport deposits follow above volcaniclastic rocks in the eastern part of the Achara-Trialeti thrust belt near Tbilisi (e.g. Beridze, 2019). Furthermore, Middle Eocene shales are exposed in the Ildokani region in the Greater Caucasus fold-and-thrust belt (Fig. 1b).
In the Kura Basin, Middle Eocene volcaniclastic rocks are overlain by up to 200 m of dark-coloured shales of the Upper Eocene Navtlugi Formation and the alternating shales, siltstones, marls and sandstones of the Tbilisi Formation (Nachtmann et al., 2015a).
Around the Eocene-Oligocene boundary, the effects of the Alpine Orogeny together with a fall in eustatic sea level resulted in separation of the Paratethys from the world ocean, leaving a semi-closed intracontinental sea that extended over large parts of central and Eastern Europe (Rögl, 1999;Popov et al., 2004b). Structural inversion of the Greater and Lesser Caucasus probably began during the Late Eocene (Vincent et al., 2007;. By contrast, the Rioni and Kura foredeeps subsided and these intermontane depressions accumulated molasse deposits from Oligocene to Neogene time (e.g. Brunet et al., 2002;Adamia et al., 2010).
Basin isolation resulted in oxygen-depleted conditions during the Oligocene and Early Miocene and the deposition of fine-grained organic matter-rich rocks with minor sandstone intervals. These largely carbonate-free pelitic rocks are referred to as the Maikop Group (Fig. 4) (e.g. Popov et al., 2004a;Sachsenhofer et al., 2017;2018b). Oligospecific nannomarls form a synchronous marker horizon in the Paratethys and correspond to a major drop in salinity at the base of nannoplankton zone NP23 (Solenovian Event; Popov et al., 2004b). In Georgia, the Maikop Group reaches a maximum thickness of 2500 to 3000 m (Popov et al., 1993).
During Middle and Late Miocene times, deepmarine marly shales and sandstones were deposited along the east-west trending basin axis and grade laterally into shallow-marine terrigenous-carbonate rocks (Adamia et al., 2011). From the Late Miocene to the end of the Pleistocene, volcanic activity occurred coeval with the deposition of molasse in the central part of the region.

PETROLEUM GEOLOGICAL OVERVIEW
Hydrocarbons are produced in Georgia in both the Rioni and Kura Basins (see Fig. 1b). The petroleum geology of these basins is summarised in the following section based mainly on previous publications (e.g. Boote et al., 2018). Source rock and petroleum geochemical data is not included because a discussion based on new data is presented in a later section.

Rioni Basin
The stratigraphic and tectonic setting of hydrocarbonproducing fields and other oil occurrences in the Rioni Basin was described recently by , a paper which forms the basis of the following summary. The location of producing oil fields and key boreholes is shown in Fig. 1b.
The structure at Supsa field is a ramp anticline forming the north-vergent leading edge of the Achara-Trialeti thrust-fold belt (Fig. 5a). The young (late Pliocene to Recent) age of the antiform was noted by Banks et al. (1997). The API of the oil in shallow (215-460 m), stacked Sarmatian sandstones varies between 22 and 29°. The deeper accumulation at the Shromisubani field is located beneath the major thrust responsible for the Supsa anticline. The oil is reservoired in Pliocene (Maeotian) clastics which are sealed by the thrust. Despite the depth (~3280 m), the oil's API is in the order of 20 to 35°. Cumulative production from both fields is about 150,000 tons (Nachtmann et al., 2015b).
The Chaladidi field contains two segments (West and East Chaladidi), and the structure is formed by ramp anticlines related to the Greater Caucasus   deformation front. Oil with 26 to 28° API has been produced from Upper Cretaceous to lower Paleocene carbonate rocks at about 2 km depth. Borehole East Chaladidi-18 also found heavy oil (12.5°) in deep (>4 km), uppermost Jurassic basaltic rocks which overlie an evaporitic succession (GGS, 2020). Some 20,000 tons of oil were produced from the Chaladidi field, but a number of similar structures (e.g. the Tsaishi, Khobi and Kvaloni anticlines; Fig. 2a) proved to lack commercial volumes of hydrocarbons (GGS, 2020). In 1975, borehole Ochamchira-1 encountered sub-commercial oil in a Middle Jurassic (Bathonian) reservoir; and in 1991, light oil was discovered at Okumi (Fig. 1b) in Upper Jurassic shallow-marine sandstones underlying Upper Jurassic evaporitic rocks that form a regional seal Fig. 2b). The high API (42.9°) and a very low Pr/n-C 17 ratio (0.1) were used to postulate a high maturity for the oil. Based on the Okumi oil discovery,  proposed the presence of two petroleum systems: a probable Cenozoic "post-salt" system, and an older "pre-salt" system.
Middle Jurassic high volatile bituminous coal at Tkibuli-Shaori (Fig. 1b), Georgia´s most important coal mine, contains a relatively high proportion of methane (~15 m³/tonne). Oil shows in Bajocian and Bathonian sandstone are common in the mine area, and >30 million m³ of thermogenic gas were produced in a single exploration well (K-1) from upper Bajocian sediments (Aslanikashvili and Iorashvili, 2013).

Kura Basin
The Upper Kura Basin is located in Georgia, whereas the lower part of the basin extends eastwards into onshore Azerbaijan. The westernmost part of the Kura Basin (to the west of Tbilisi) is also known as the Kartli Basin, and includes a 4 to 5 km thick Paleogene to Quaternary succession (Banks et al., 1997;Fig. 2b). Its southern margin is formed by the frontal part of the Achara-Trialeti thrust-fold belt (Alania et al., 2017). Oil has been encountered in several small fields including Uplistsikhe-5 (light oil with API 41.3°), Akhalkalaki and Kavtiskhevi Fig. 5c). At Akhalkalaki, oil shows were detected in the 1970s in Upper Eocene deposits (well Akhalkalaki-8) at depths between 550 and 800 m and at even shallower depths (400-500 m) in the Upper Cretaceous (well Akhalkalaki-7). The most important oil accumulations, however, are located in the Tbilisi region (Fig. 1b). By far the largest oil field in Georgia is Samgori (discovered in 1974), which is accountable for 24.2 million tons or 87 % of the oil produced in the country (Nachtmann, 2015b). The Samgori field is about 24 km long and comprises the Samgori, Patardzeuli and Ninotsminda structures (Fig. 5d), which are located on a single anticline but whch represent three discrete structural closures (Patton, 1993). A permeability barrier, possibly a strike-slip fault, is present between the Patardzeuli and Ninotsminda sectors. The maximum structural closure is 760 m on the Ninotsminda culmination. Oil mainly occurs in fractured Middle Eocene volcaniclastic reservoir rocks. Middle Eocene oils are low sulphur (0.22 %) with an average API gravity of 39.2°. Minor quantities of a similar oil have been produced from Upper Eocene sandstones, and dry gas (~98 % CH 4 ) is produced from Lower Eocene sandstones. At Ninotsminda, the gas cap is about 240 m thick. Hydrodynamic influences are indicated by the inclined oil-water contact (~1°) and the low salinity (4.5-6.5 mg/l) of the formation water (Patton, 1993). Considering the north-directed foreland deformation of the Achara-Trialeti thrust-fold belt, the domal structure is probably formed by inversion-related anticlines and pop-up -like structures riding over high-angle reverse faults (Pace and GOG, 2019).
Other oil fields in the Tbilisi area producing from Middle Eocene volcaniclastics include South Dome (discovered in 1973; cumulative production 1.15 million tons), Teleti (540, 000 tons) and West Rustavi (1987; <100,000 tons). Gas and oil are produced at Rustavi. The Norio (1938) and Satskhenisi (1956) fields are located on the same anticlinal trend and produce oil from steeply dipping Lower Miocene (Satskhenisi) and Middle/Upper Miocene sandstones (Norio) at depths between 500 and 2000 m (cumulative production ~300,000 tons). Oil has also been reported in Upper Cretaceous fractured limestones in anticlinal structures (e.g. Kavtiskhevi, Manavi; Nachtmann et al., 2015a), and in Lower Cretaceous turbidites in the Greater Caucasus fold-and-thrust belt (Vedzebi North; 30.4°; Pr/Ph: 1.99; Robinson et al., 1997), but commercial production has not yet been established from Mesozoic rocks.

Samples
Nineteen oil and five gas samples, mostly provided by Georgia Oil & Gas (GOG), were analysed in the present study at Montanuniversitaet Leoben. Five oil samples came from producing fields in the Rioni Basin (Supsa, Shromisubani, East Chaladidi), and their analyses is discussed together with data from two Shromisubani oils studied by Mayer et al. (2018). One additional oil sample came from the Racha Shaori coal mine.
Two oil samples came from Akhalkalaki in the Kartli Basin west of Tbilisi, and four oil samples were from Ninotsminda. Amongst these, only sample Ninotsminda-98 was from the main Middle Eocene reservoir, whereas samples Ninotsminda-21, -59 (Upper Eocene) and -78 (Oligocene) were from subordinate reservoirs. Another five oil samples came from producing fields in the Tbilisi area (Satskhenisi, Teleti and Rustavi); one sample was from a noncommercial oil discovery in a Cretaceous reservoir (Manavi-12); and one sample was from a hydrocarbon seep (Kvelatsminda). Taribani-39 was the only sample from eastern Georgia. All five gas samples were from producing fields in the Tbilisi area. The stratigraphic position of the reservoir rocks of oil and gas samples is shown in Fig. 4.
A total of 165 cuttings samples of Eocene and Oligocene (Maikopian) organic-rich sediments were collected in boreholes located to the east (Pat-E1) and SE of Tbilisi (Rustavi-16, Kumisi-2). Another 100 samples with Early and Middle Miocene ages from the Intsra outcrop section (Fig. 1b) were provided by D. Palcu. GOG currently undertakes systematic sampling of all potential organic-rich units in the Jurassic to Paleogene succession in Georgia, and about 200 surface samples have so far been collected. However, a significant number of these samples contain very low TOC contents or are thermally overmature. The present paper therefore only focuses on a number of the most relevant samples. These were derived from the northern margin of the Rioni Basin (Enguri-Kutaisi-Oni) and from the Eocene succession to the south and west of Tbilisi and in the Ildokani area (see Fig. 1b). The new source rock data is discussed below together with data published by Pupp et al. (2018).

Methods
Oil samples: The oil samples were dissolved in n-hexane, followed by separation of the precipitated asphaltenes from the solution by centrifugation. The hexane-soluble fractions were separated into NSO-compounds, saturated hydrocarbons and aromatic hydrocarbons using medium-pressure liquid chromatography (MPLC) with a Köhnen-Willsch instrument (Radke et al., 1980). The saturated and aromatic hydrocarbon fractions were analysed using a gas chromatograph equipped with a 60 m DB-5MS fused silica capillary column (i.e. 0.25 mm; 0.25 µm film thickness) coupled to a ThermoFisher ISQ quadrupole mass spectrometer. The oven temperature was programmed from 40° to 310°C at 4°C min -1 , followed by an isothermal period of 30 min. Helium was used as the carrier gas. Samples were injected splitless, with the injector temperature at 275°C. The spectrometer was operated in the EI (electron ionisation) mode over a scan range from m/z 50 to m/z 650 (0.5 s total scan time). Data were processed with an Xcalibur or Chromeleon data system. Individual compounds were identified on the basis of retention time in the total ion current (TIC) chromatogram and by comparison of mass spectra with published data. Relative percentages and absolute concentrations of different compound groups in the saturated and aromatic hydrocarbon fractions were calculated using peak areas in the TIC chromatograms in relation to those of internal standards (deuteriated n-tetracosane and 1,1´-binaphthyl, respectively), or by integration of peak areas in appropriate mass chromatograms using response factors to correct for the intensities of the fragment ion used for quantification of the total ion abundance.
The n-alkanes were separated from branched/ cyclic hydrocarbons by an improved 5 Å molecular sieve method (Grice et al., 2008) for the analysis of stable carbon isotope ratios on individual n-alkanes and isoprenoids. Stable C isotope measurements were performed using a Trace GC-ultra gas chromatograph attached to the ThermoFisher Delta-V isotope ratio mass spectrometer (IRMS) via a combustion and high temperature reduction interface, respectively (GC Isolink, ThermoFisher). The GC coupled to the IRMS was equipped with a 30 m DB-5MS fused silica capillary column (i.d. 0.25 mm; 0.25 µm film thickness). The oven temperature was programmed from 70 to 300°C at a rate of 4°C/min followed by an isothermal period of 15 min. Helium was used as the carrier gas. The sample was injected splitless at 275°C. CO 2 monitoring gas, calibrated versus the PeeDee belemnite standard (VPDB) by the NBS-19 reference material, was injected at the beginning and at the end of each analysis. Isotopic compositions are reported in δ notation relative to the V-PDB standards. Analytical reproducibility (0.2‰ for δ 13 C) was controlled by repeated measurements of n-alkane standard mixtures.
Gas Samples: Molecular gas compositions were determined on a ThermoFisher Trace GC-ultra equipped with three individual gas channels. Hydrocarbons were resolved on a 30 m Rtx-Alumina capillary column (i.d. 0.53 mm; filling Na 2 SO 4 , 10 µm film thickness) and detected by a flame ionization detector (FID). O 2 , CO 2 and N 2 were resolved on two packed columns: HayeSep Q (2 m x 1/8" OD) and MolSieve 5A (2 m x 1/8" OD) and detected by TCD. Sulphur compounds were determined using a Rtx-Sulfur packed column (2 m x 1/8" OD) and a flame photometric detector (FPD). The column oven was programmed to hold at 50°C for 260 seconds and then to increase to 165°C at 10°C/min, at which point it was held for 30 seconds. Helium was used as the carrier gas for all three channels.
Stable C and H isotope measurements were carried out using a Trace GC-ultra gas chromatograph attached to a ThermoFisher Delta-V IRMS via a combustion and high-temperature reduction interface, respectively (GC Isolink, ThermoFisher). The GC coupled to the IRMS was equipped with a 25 m PoraPlot capillary column (i.d. 0.32 mm; 0.10 µm film thickness). The oven temperature was programmed from 30 to 180°C at 5°C/ min followed by an isothermal period of 5 min. Helium was used as the carrier gas. For calibration, a CO 2 or H 2 monitoring gas was injected at the beginning and end of each analysis. Analytical reproducibility was controlled by repeated measurements of the calibration gas.
Rock samples: Total carbon (TC), total sulphur (S) and total organic carbon (TOC) contents were analysed using an ELTRA Elemental Analyser. Samples for TOC measurements were decarbonized with concentrated phosphoric acid. TC and TOC were used to calculate calcite equivalent percentages ([TC-TOC]*8.333). Pyrolysis measurements were performed using a Rock-Eval 6 instrument. The S1 and S2 peaks (mg HC/g rock) were used to calculate the petroleum potential (S1+S2), the production index (PI = S1/(S1+S2)) and the hydrogen index (HI = 100*S2/TOC mg HC/g TOC).
The temperature of maximum hydrocarbon generation during pyrolysis (T max ) was used as a maturity indicator. For biomarker and isotope analysis, rock samples were extracted for ca. 1 h using dichloromethane in a Dionex ASE 350 accelerated solvent extractor at 75 °C and 75 bars. After evaporation of the solvent in a Zymark TurboVap 500 closed cell concentrator, asphaltenes were precipitated from a hexane-dichloromethane solution (80:1) and separated using centrifugation. Separation and analyses of the hydrocarbon fractions were obtained in the same way as oil samples.
Hierarchical cluster analysis was performed to group the oil samples and for source-to-oil correlations, and was completed using autoscale preprocessing, Euclidean metric distance and the farthest neighbour option. The analysis included source-related biomarker and isotopic parameters (e.g. Pr/Ph ratios, sterane distribution, compound-specific isotope data of n-alkanes and isoprenoids).

Oil geochemistry
Geochemical data for oil samples from the Supsa, Shromisubani and East Chaladidi fields and for liquid bitumen recovered from the Racha Shaori coal mine ( Fig. 1b) are presented in Table 1 (page 312).
The shallow Supsa oil (~550 m) is very rich in NSO compounds (61 %) and lacks n-alkanes and isoprenoids. This shows that the Supsa oil reached level 4 of the biodegradation scale of Peters and Moldowan (1993). The deeper Shromisubani oil (~3000 m) is less strongly affected by biodegradation (level 2-3), but pristane/n-C 17 and phytane/n-C 18 ratios are elevated. The East Chaladidi oils and the Racha Shaori bitumen are not significantly biodegraded, although n-alkanes with less than 15 C-atoms are depleted in the latter, possibly due to evaporative loss.
Pristane/phytane (Pr/Ph) ratios are lower in the Racha Shori bitumen (1.25) and East Chaladidi oils (1.34-1.42) than in the Shromisubani oils (1.62-1.72). Highly-branched isoprenoid (HBI) alkanes were detected in quantifiable amounts only in the Shromisubani oils. Fig. 6 shows that the relative percentages of C 27 , C 28 and C 29 steranes are similar in the Supsa, Shromisubani and East Chaladidi oils suggesting that sterane distributions in the Supsa oil are not affected by biodegradation. However, the Racha Shaori bitumen is strongly depleted in C 28 steranes. Consequently, C 28 / C 29 sterane ratios are higher in the Supsa, Shromisubani and East Chaladidi oils (0.74-0.98) than in the Racha Shaori bitumen (0.33; Table 1). The ratios of C 27 diasteranes to the corresponding regular steranes vary strongly (0.36-0.90). The highest ratio was observed in the Shromisubani oil (Table 1).
Oleanane, an angiosperm-derived triterpane (Riva et al., 1988), was found in significant amounts in the Supsa, Shromisubani and East Chaladidi samples, in which the oleanane index varies between 0.13 and 0.19 (Table 1). This suggests a Cenozoic (or Upper Cretaceous) source rock for these oils. In contrast, oleanane was not detected in the Racha Shaori bitumen which, together with the low C 28 /C 29 steranes ratio, strongly indicates a pre-Upper Cretaceous source rock (c.f. Grantham and Wakefield, 1988).
The DBT/Phen ratio is low for all the oils (0.7-0.9), showing that the amount of free H 2 S in the water column at the time of source rock deposition was limited (Hughes et al., 1995). Hence, it is not surprising that gammacerane was not detected in any of the oil samples. Aryl-isoprenoids were present in the Shromisubani and East Chaladidi oils in varying concentrations (Table 1).
The C 31 hopane 22S/(22S+22R) isomerization ratios (0.57-0.60) are close to the end-point value (Mackenzie and Maxwell, 1981), while the C 29 aaa sterane 20S/(20S+20R) ratio (0.31-0.49) did not reach the end-point value of 0.55 (Mackenzie and Maxwell, 1981). Ratios of abb /(abb+aaa) C 29 -steranes range from 0.42 to 0.54. Fig. 7 shows that the Racha Shaori bitumen was generated at peak oil maturity, while the Supsa, Shromisubani and East Chaladidi oils were generated in the early oil window. This assessment is supported by the methylphenanthrene index (MPI 1; C S te r a n e ( % ) 2 8 C S t e r a n e ( % )  Radke and Welte, 1983), which ranges from 0.35 to 0.60 (Table 1) suggesting that the aromatic compounds in the Supsa, Shromisubani and East Chaladidi oils were generated between 0.61 and 0.76 % vitrinite reflectance equivalent (R oe ). Isotope ratios of n-alkanes are listed in Table 2 (page 313) and are plotted versus chain length in Fig. 8. Oils from different locations show significant differences: short-chain n-alkanes in oils from the East Chaladidi field are isotopically heavier and long-chain n-alkanes isotopically lighter than Shromisubani oils. In contrast to oils in Cretaceous and Cenozoic reservoirs, n-alkanes in the Racha Shaori bitumen are isotopically very light (~ -32.5‰; Fig. 8).
Source Rocks "Post-salt" source rocks Detailed data for Eocene and Oligocene source rocks in the Rioni Basin were presented by Pupp et al. (2018) and are summarized below. Profiles of geochemical  parameters and isotope ratios versus depth for the Martvili outcrop section (for location see Fig. 1b) are shown in Fig. 9 ; here the Middle Eocene (Bartonian) Kuma Formation and the Oligocene part of the Maikop Group are exposed. The Kuma Formation is widespread in the Rioni Basin where it is referred to locally as the 'Lirolepisic' sequence, but is replaced by volcanogenic rocks in the southern part of the basin . The formation is about 40 m thick and is typically composed of marls. In outcrop sections, the organic matter is immature. The average TOC is 3.2 wt.%, and hydrogen indices (HIs) of 300-600 mg HC/g TOC indicate oil-prone Type II kerogen (Fig. 10a). The petroleum potential classifies the Kuma Formation as a "good" source rock (Fig. 10b). The oil generation potential is between 1.0 and 2.4 t HC/m² (Pupp et al., 2018) showing that the Kuma Formation is a rich source rock. Pr/Ph ratios are typically slightly above 1.0. C 25 HBI thiophenes and aryl-isoprenoids are present in varying concentrations. C 27 , C 28 and C 29 steranes occur in similar proportions (Fig. 6). Compound-specific isotope data indicate that longchain n-alkanes have δ 13 C values which are slightly less negative than short-chain n-alkanes (Fig. 11).
The Maikop Group lies above the Upper Eocene Belaya Glina Formation (Fig. 9) and is divided into Pshekhian, Solenovian-to-Kalmykian, and Lower Miocene intervals. At the Martvili section, the Pshekhian interval (NP21-22) is more than 60 m thick and comprises a marly lower part and a shale-rich upper part, and contains high quantities (average 2.7 wt.% TOC) of Type II-III kerogen (average HI: 278 mg HC/g TOC). Calcareous shales deposited during the Solenovian Event (at the onset of nannoplankton zone NP23) form the base of the largely carbonatefree Solenovian-to-Kalmykian shale shale succession. This interval, 424 m thick, is less organic-rich (~2.0 wt.%TOC) and contains dominantly Type III kerogen (average HI: 140 mg HC/g TOC). The Lower Miocene part of the Maikop Group, studied at the Intsra outcrop section (see Fig. 1b  . o h t i L P N . t a r t S carbonate-free, but contains even lower quantities (~1.3 wt.%TOC) of Type III kerogen (average HI: 90 mg HC/g TOC; see also Vincent and Kaye, 2018). The higher petroleum potential of the Pshekhian succession is also suggested by a plot of petroleum potential versus TOC (Fig. 10b). In total the Maikop Group has a generation potential of about 4 t HC/m², a value which is higher than in most other sub-basins in the Eastern Paratethys (Sachsenhofer et al., 2018a, b). Pr/Ph ratios are above 1 near the base of the Oligocene succession but decrease up-section to values significantly below 1. HBI thiophenes and aryl-isoprenoids are present.
Relative percentages of C 27 , C 28 and C 29 steranes are similar to those in the Kuma Formation (Fig. 6). In comparison to the Kuma Formation, short-chain n-alkanes contain carbon with slightly less negative d 13 C values (Fig. 11a).
The Kuma Formation and the Maikop Group include proven high quality source rocks, but additional source intervals may also be present.  proposed that the lower part of the Middle Miocene succession ("Tarkhanian shales") may have source potential, but data from the Intsra section show that both Tarkhanian and Chokrakian shales are poor source rocks (Fig. 10b).
Some Lower Cretaceous (Albian) rocks exposed at the northern margin of the Rioni Basin (near the Enguri dam; Fig. 1b) show good potential (Fig. 10). Samples of the Lower Cretaceous succession from locations to the west of Oni (Racha in Fig. 10a) may have high TOC contents with Type II-III kerogen and good petroleum potential. However, some of these samples have recently been re-interpreted as Paleogene in age (Coric, pers. comm., 2020). Biomarker or isotope data for these rocks are not yet available.
"Pre-salt" source rocks Little data on potential "pre-salt" source rocks are available. Robinson et al. (1997) speculatively proposed a Kimmeridgian source rock but this was not supported by any geochemical data. Toarcian rocks with TOC contents between 0.5 and 2.8 wt.% TOC were mentioned by Yükler et al. (2019) but more detailed data are absent.
Middle Jurassic rocks have been investigated in the context of the present study. The analysed Aalenian shale sample from a location east of Oni is overmature (T max : 468°C) and contains 1.2 wt.% TOC. Because of its high maturity, the HI is low (56 mg HC/g TOC). Bajocian samples from near Enguri are marginally mature, and have varying TOC contents (max. 1.5 wt.%) but low HI values (24-63 mg HC/g TOC) despite relatively low maturity (Fig. 10a).
Bathonian shales about 300 m thick, interpreted as lacustrine, are exposed near the Mtischala River (Fig.  1b). These rocks are marginally mature, and three samples from the lower part of the section have very high TOC contents (1.8-14.1 wt.%) with Type III to II-III kerogen (HI 172-276 mg HC/g TOC; Fig. 10a) and fair to very good petroleum potential (Fig. 10b). The kerogen type indicates a significant input of terrigenous organic matter, which is consistent with the presence of abundant fossil driftwood observed in the section. At other locations (e.g. Kursebi; Fig. 1b), Bathonian sediments are also organic-rich but are mature or even overmature, and are therefore characterized by low  HI values and a low remaining petroleum potential (grey area in Fig. 10b). The high maturity is at least partly caused by heating associated with volcanic sills of unknown age. In any case, Bathonian sediments may be an important source rock for gas and oil. High gas potential is also supported by the presence of commercial coal seams (Adamia et al., 2011).

Oil geochemistry
For oils samples analysed from fields in the Kura Basin, there are major differences in the relative concentrations of hydrocarbon fractions (Table 5: page 314). Partly, this indicates different degrees of biodegradation, as also suggested by the Pr/n-C 17 and Ph/n-C 18 ratios (Fig. 12). Thus, oil seeping out of Mesozoic rocks near Kvelatsminda is severely biodegraded (level 5 after Peters and Moldowan, 1993). Oil produced from Miocene reservoir sandstones in the Satskhenisi-3 well reached level 2 on the biodegradation scale, although the specific gravity (41.0-53.6°API; Khetsuriani et al., 2016) and biomarker data from internal industry reports suggest that the oil from Satskhenisi is only slightly biodegraded if at all (level 0-1). In contrast, unpublished reports suggest that oils from Norio are strongly biodegraded as shown by the removal of n-alkanes and isoprenoids (level 3-4). In the Ninotsminda field, oils from Middle Eocene reservoirs have lower Pr/n-C 17 and Ph/n-C 18 ratios than oils from Upper Eocene and Oligocene reservoirs.
Pr/Ph ratios are generally high (≥1.5): the average ratio for four Ninotsminda oils is 2.2, and even higher Pr/Ph ratios are observed in the oils from Rustavi (2.9-3.6) and Akhalkalaki (2.3-2.8; Table 5). HBI alkanes occur in oils from Manavi and Ninotsminda, and in one of the Teleti oils; the highest amounts occur in the severely biodegraded Kvelatsminda seepage oil. The oil samples typically contain slightly higher relative percentages of C 27 steranes than C 29 and C 28 steranes, but sterane distributions are fairly uniform (Fig. 13). C 28 /C 29 sterane ratios are generally high and range from 0.78 to 1.14 ( Table 5). Diasterane/sterane ratios (0.34-0.98) vary considerably and are high (≥ 0.9) in the oils from Akhalkalaki, Manavi, Rustavi and Taribani.
Oleanane is present in all the oils analysed from the Kura Basin. The oleanane index varies between 0.11 and 0.29 (Table 5) and is often higher (>0.2) than in oils from the Rioni Basin. Oleanane is also present in the Kvelatsminda seepage oil and in the Akhalkalaki 7 oil which is produced from an Upper Cretaceous reservoir. Hence a Cenozoic (or Upper Cretaceous) source rock is very likely for all the oils.
All the oils are characterized by very low DBT/ Phen ratios (0.01-0.17), which are lower than the ratios for Rioni Basin oils. Gammacerane has not been recorded in the Akhalkalaki and Taribani oils but is present in most of the oils analysed from the Tbilisi region (Table 5).
C 31 hopane 22S/(22S+22R) isomerization ratios (0.57-0.61) are close to the end-point value (c.f. Mackenzie and Maxwell, 1981). The C 29 aaa sterane ratios of 20S/(20S+20R) isomers typically range from 0.39 to 0.56. A maximum value of 0.67 was observed in the severely biodegraded Kvelatsminda oil sample and probably reflects the preferential biodegradation of the 20R epimer. Ratios of abb/(abb+aaa) C 29 -steranes range from 0.51 to 0.69. Fig. 7 shows that the oils in the Kura Basin were generated at peak oil maturity and therefore at a higher maturity than the oils in the Rioni Basin. MPI 1 values (0.54-0.75) suggest that the aromatic fractions of the Kura Basin oils were generated at 0.72 %R oe (Taribani) to 0.85 %R oe (Akhalkalaki).
Plots of d 13 C versus chain length of n-alkanes show minor but systematic changes (Table 6, page 315; Fig. 11). Akhalkalaki oils (Kartli Basin) show rather uniform, relatively heavy carbon isotope values with a local minimum at n-C 26 . The remaining oil samples show a continuous decrease of d 13 C with increasing chain length. For oils from the Tbilisi region, a NE-SW trend of decreasing d 13 C values is apparent. The least negative values were observed in the Taribani oil from eastern Georgia.

Gas geochemistry
The molecular and isotopic compositions of three gas samples from the Samgori-Patardzeuli-Ninotsminda field and one sample from the West Rustavi field were determined (Tables 7, 8: page 315). The results are displayed in discrimination plots after Bernard et al. (1977) and Whiticar (1999), updated by Milkov and Etiope (2018;Fig. 14). As expected, all samples plot in the field of thermogenic gas (Fig. 14). Gas dryness and carbon and hydrogen isotope data suggest

Fig. 12. Pristane/n-C 17 versus phytane/n-C 18 ratios (after Connan and Cassou, 1980). Data from Uplistsikhe and Vedzebi oils (yellow squares) are from Robinson et al. (1997). The fields for Rustavi and Ninotsminda oils include data from unpublished industry reports.
a maturity trend, with Nonitsminda gas from Upper Eocene reservoirs showing the lowest maturity and Samgori gas from Lower Eocene reservoirs the highest maturity. In the original version of the Bernard plot (Bernard et al., 1977), the gas samples plot in the field characteristic of Type III kerogen. However, Milkov and Etiope (2018) emphasized that the larger dataset used in their study does not allow gases generated by Type II or Type III kerogens to be distinguished. In a "Natural Gas Plot" (Chung et al., 1988;Fig. 15), carbon isotopic compositions of individual hydrocarbons are plotted as a function of carbon number. Samples generated by primary cracking of source kerogens should lie on a straight line when plotted against the inverse of the carbon number, and the gradient of the line reflects differences in thermal maturity. Hence, the high gradient of the Upper Eocene Ninotsminda gas supports its low maturity (Fig. 15). However, a small admixture of isotopically light methane in samples from Satskhenisi and Ninotsminda (Upper Eocene) cannot be excluded. Robinson et al. (1997) considered organic-rich shales (max. TOC 4.3 wt.%; max. HI: 416 mg HC/g TOC), reportedly of Late Eocene age, to be the main source rock in the Kura Basin. However, the age of these rocks still needs verification (e.g. Boote et al., 2018). In addition, the Maikop Group is widely considered to be an important source rock in the Kura Basin. However, Pupp et al. (2018) showed that the Maikop Group in the Tbilisi area has relatively poor source rock characteristics. Hence, the quality of the Maikop Group is re-assessed in this section and additional potential rock intervals are explored.

Maikop Group
Outcrop samples of the Maikop Group from the east of Tbilisi have moderately high amounts of organic matter but low HI values (Pupp et al., 2018;Fig. 16a) and their petroleum potential is therefore classified as poor (Fig. 16b). As higher HI values have been recorded in cuttings samples from some wells, these samples were re-analysed. Pyrograms suggest that HI values exceeding 220 mg HC/g TOC may have been caused by contamination in boreholes drilled with oil-based mud. A geochemical log of the upper part of borehole PAT-E1 drilled in the Patardzeuli area (Korelskiy et al., 2019;Fig. 17) supports the poor (to fair) petroleum potential of the Maikop Group (Figs 16a, b).
Pr/Ph ratios of Maikop samples from the Tbilisi area are typically high (0.6-5.3; average: 2.0; Pupp et al., 2018). Biomarker data for two Maikop samples are listed in Table 9 (page 316) and support the high Pr/Ph ratios. Isotope patterns of n-alkanes in the extracts of these two samples (Table 10, page 316) are characterized by a decrease in δ 13 C between n-C 15 and n-C 23 and moreor-less constant values up to n-C 29 (Fig. 11b).

Tbilisi Formation
The Upper Eocene Tbilisi Formation underlies the Maikop Group. In the PAT-E1 borehole, the Tbilisi Formation has low TOC contents and contains Type III kerogen (Figs 16, 17) and it is therefore not considered to be a potential source rock (Fig. 16b).

Navtlugi Formation
The lower part of the Upper Eocene succession consists of dark-coloured fine-grained rocks assigned to the Navtlugi Formation, which form the seal for the Middle Eocene volcaniclastic reservoir unit in the Tbilisi area.
The hydrocarbon potential of the Navtlugi Formation was investigated using 59 cuttings samples from boreholes PAT-E1, Kumisi-2 and West Rustavi-16, where the formation is marginally mature (average T max ~440°C).
In the PAT-E1 borehole, the Navtlugi Formation is about 200 m thick and has on average 1.5 wt.% TOC (Fig. 16). The average HI is slightly less than 200 mg HC/g TOC. Higher HI values (average 220; max. 320 mg HC/g TOC) were observed in the cuttings from borehole Kumisi 2, where TOC contents increase downwards from 1.0 to 2.0 wt.% in a 100 m thick interval. In Rustavi-16, the Navtlugi Formation is about 200 m thick and has on average 1.6 wt.% TOC (1.3-2.2 wt.%) with an average HI of 202 mg HC/g TOC. Overall, the petroleum potential of the Navtlugi Formation is classified as fair (Fig. 16b). Compoundspecific isotope patterns from two Navtlugi samples from borehole Kumisi-2 are shown in Fig. 11b, and are characterized by a specific V-shaped pattern with a minimum at n-C 19 .

Middle Eocene
In the easternmost part of the Achara-Trialeti foldthrust belt, volcaniclastic rocks are overlain by upper Middle Eocene submarine mass-transport deposits (Beridze, 2019). Fine-grained rocks interfingering with the mass-transport deposits were investigated in outcrops south and west of Tbilisi (Botanic Garden, Didgori, Kumisi). At Kumisi, these rocks are immature and contain moderate amounts of organic matter (TOC 1.1-2.5 wt.%) with HI values ranging from 150 to 207 mg HC/g TOC (Fig. 16a,b). The remaining petroleum potential of these rocks at other localities is lower (Fig.  16b) because of their higher maturity.
At Ildokani in the Greater Caucasus north of Tbilisi, the Middle Eocene succession includes mudstones with organic-rich layers (2.7 -5.3 wt.% TOC) dated as Bartonian to earliest Priabonian. These layers contain Type II to III kerogen (HI 222-292 mg HC/g TOC; Fig. 16a) and therefore have higher petroleumgenerating potential than any other stratigraphic interval investigated in the Kura Basin (Fig. 16b). Isotope data from Middle Eocene rock extracts (Table  10) show a continuous decrease in δ 13 C with increasing chain length, and the gradient is higher in samples from Ildokani than in those from Tbilisi (Fig. 11b).

Lower Eocene
Lower Eocene rocks were drilled beneath the volcaniclastics in the PAT-E1 well. These rocks are characterized by TOC contents less than 1.0 wt.% and do not have any petroleum potential (Fig. 16b).

Oil families and oil-source correlation
In spite of the systematic investigation of source rocks and accumulated oils in the Rioni and Kura Basins, oil-source correlation is still somewhat ambiguous. Current ideas supported by hierarchical cluster analysis are summarized below:

Rioni Basin
Supsa, Shromisubani: Based mainly on compoundspecific isotope data, Mayer et al. (2018) postulated that the Shromisubani oil is a mixture of oil generated from Middle Eocene Kuma and Oligocene Maikop source rocks, and that the oil has migrated laterally from deeper parts of the basin in the offshore Black Sea area. In general, this interpretation is still valid although oil generation may also have occurred in synclinal structures to the north and south of the Supsa/Shromisubani structure. However, the Kuma Formation is probably not present in the southern Rioni Basin (see Fig. 4). Because of the advanced levels of biodegradation (level 4), compound-specific isotope data for the Supsa oil are not available. However, isotope data from oil fractions show that the Supsa oil is isotopically heavier (saturates: -26.0‰; Robinson et al., 1996) than the Shromisubani oil (saturates: -28.1‰; Robinson et al., 1997). A possible explanation for this discrepancy is biodegradation that resulted in an increase in the δ 13 C values of the saturate fraction (Pedentchouk and Turich, 2017). Moreover, similarities in biomarker data suggest a common source for the Supsa and Shromisubani oils. Robinson et al. (1997) investigated oil seeping from the well-head of borehole Natanebi-2 (site I in Fig. 1b) and found that the oil had Pr/Ph ratios and relative amounts of steranes and isotope ratios like those in the Shromisubani oil. This indicates that all the oils from the hanging wall of the Achara-Trialeti thrust-fold belt have the same Cenozoic source.
East Chaladidi: The East Chaladidi oils were generated at the same low maturity as the Supsa and Shromisubani oils (Fig. 7). Differences in biomarker proxies are minor, but compound-specific isotope patterns are clearly different (Fig. 8) suggesting the presence of a separate oil family. The presence of oleanane shows that at least some of the oil has an Upper Cretaceous or Cenozoic source. In this regard, the deep East Chaladidi-18 oil (~4.4 km), reportedly recovered from Upper Jurassic basalts, is identical to shallower East Chaladidi-3 and -13 oils (~2.0 km) and shows the same low maturity (Fig. 7). Hence there are doubts as to whether the former sample was really recovered from Upper Jurassic basalts, and in any case the source of the East Chaladidi oils is problematic.  Okumi, Racha Shaori and the question of a pre-salt petroleum system: The absence of oleanane, low C 28 / C 29 steranes ratios, n-alkanes with very negative d 13 C values (~ -33‰), and the geological setting suggest that the Racha Shaori bitumen was generated from Bathonian organic-rich shales. This provides additional evidence for the presence of a working "pre-salt" petroleum system as postulated by  based on the Okumi oil discovery in Upper Jurassic sandstones.
However, the origin of the Okumi oil remains enigmatic. Because this oil is no longer accessible, the following discussion is based on partly inconsistent literature data. In contrast to Robinson et al. (1996), a later study by the same author  reported the presence of oleanane but could not exclude some form of contamination. Moreover, apart from the apparent presence of oleanane, the relatively high content of C 28 steranes (Fig. 6) and the high C 28 /C 29 sterane ratio (0.84) are arguments against a Bathonian or Aalenian source (Fig. 6). In addition, the high Pr/Ph ratio (1.9) reported by Robinson et al. (1997) also makes a Bathonian source rock less likely. Furthermore, bulk parameters presented above show that the Bajocian succession cannot be considered a source rock. Robinson et al. (1997) showed that the saturated fraction (δ 13 C: -31.9 ‰) is isotopically significantly lighter than Kuma or Maikop extracts. Summarizing, the Okumi oil is not comparable to any other oil from Georgia and forms a separate oil family. Its origin remains problematic, and detailed biomarker and isotopic investigations of Cretaceous to Paleogene source rocks along the northern margin of the Rioni Basin will be required to identify the source rock.
The results of cluster analysis (Fig. 18a) support the above considerations. Notably, they suggest that there is a closer relationship between the Shromisubani and East Chaladidi oils and the Kuma Formation than between these oils and the Maikop Formation. The Racha Shori bitumen and the Bathonian shale form a distinct cluster.

Kura Basin
Oleanane is present in all the Kura Basin oils, including the oils produced from Cretaceous units at boreholes Vedzebi North-1 (Robinson et al., 2017), Akhalkalaki and Manavi. This suggests a Cenozoic or Upper Cretaceous source rock. Typically, the oils display a higher thermal maturity (peak oil) than the Rioni Basin oils (Fig. 5). Yukler et al. (2019) suggested that this may be a result of delayed secondary migration in the Kura Basin. Potential carrier beds are of relatively poor quality, and hydrocarbons therefore remained in the kitchen area until strike-slip faults triggered by thin-skinned thrusting enabled secondary migration to occur.
Pr/Ph ratios are generally high with maximum Pr/ Ph values in the oils from Rustavi field (3.22) and the Kartli Basin (Akhalkalaki-7, -8, Uplistsikhe-5: 2.60; note erroneous Pr/Ph value in Robinson et al., 2017). Considering unpublished data from internal reports, the average Pr/Ph ratios of Ninotsminada oils from Upper Eocene and Oligocene reservoirs (2.4) appear to be higher than those of oils from the main Middle Eocene reservoir (2.0). Therefore, the data suggest a source rock with a significant but varying contribution of land plant material. In fact, all the relevant Middle Eocene to Oligocene source rocks are characterized by high Pr/ Ph ratios (Table 9) and cannot be excluded as potential source rocks on that basis. Differences in the sterane distributions of the oils are minor. Only the Vedzebi North-1 oil exhibits an unusual low percentage of C 27 steranes. The best match is with Middle Eocene rocks in the eastern Achara-Trialeti thrust-fold belt (Tbilisi area) and Maikop samples. In contrast, both Navtlugi samples have relatively low C 28 sterane contents. The fit between oil samples and Middle Eocene samples from the Ildokani region is likewise unsatisfactory.
Important additional information is provided by compound-specific isotope patterns. Akhalkalaki oils show a rather uniform isotope distribution whereas all other oils show a decrease in δ 13 C ratios with increasing chain length. This suggests the presence of two oil families. In oils from the Tbilisi region, the δ 13 C values increase from northeast to southwest, probably reflecting the increasing maturity of the source rock in the same direction (Pedentchouk and Turich, 2017). Based on this consideration, the isotopically heavy Taribani oil may have been generated at an even higher maturity. Although there is a weak trend in saturated-based maturity parameters supporting the southwestward increase of maturity for the Tbilisi region oils (Figs 7, 12), a clear trend is missing. Alternative interpretations include lateral facies variations within a single source rock, or mixing of oils from two (or more) source rocks. This concept agrees with models of migration pathways in the Tbilisi area developed by Yukler et al. (2019), who postulated a northern oil kitchen which successively charged the Samgori-Patardzeuli-Ninotsminda, South Dome, Teleti and Rustavi fields. For the latter field, the authors assumed an additional charge from the SE which, based on cluster analysis, may be derived from Maikop or Eocene shales (Fig. 18b).
Compound-specific isotope patterns of source rocks show that the Navtlugi Formation, which is characterized by a specific V-shaped pattern, somewhat surprisingly did not contribute significantly to the accumulated oil (see also Fig. 18b). In addition, the very light d 13 C values in long-chain n-alkanes (Fig.  11b) suggests that the Middle Eocene in the Ildokani region, the best source rocks in eastern Georgia, likewise did not contribute significantly (see also Fig.  18b). In contrast, there is a relatively good correlation between compound-specific isotope patterns of the oils and the Maikop samples. Within this context, it should be noted that isotope patterns of Maikop Group sediments in the Kura Basin vary significantly both laterally and vertically, as is shown by isotope data from Lower Oligocene and Upper Oligocene Maikop sediments in the Lower Kura Basin, Azerbaijan (see Fig. 11). Hence, no clear oil-source correlation can yet be performed. However, the Maikop Group is the most likely source rock, despite its low petroleum potential. Some additional oil may have been contributed by Middle Eocene rocks in the eastern part of the Achara-Trialeti fold-thrust belt.

Definition of Petroleum Systems
Petroleum systems events charts have been compiled for "pre-salt" and "post-salt" petroleum systems in the Rioni Basin and for the petroleum system in the Kura Basin (Fig. 19). A common feature of these petroleum systems is the relatively recent timing of hydrocarbon generation. Thus the preservation time is very short and the critical moment is very recent. Recent hydrocarbon generation is also reflected by abundant surface seeps (Nibladze and Janiashvili, 2014), which often contain oil with only minor biodegradation.

"Pre-salt" petroleum system(s) in the Rioni Basin
Bathonian shales are the only proven source rocks in the speculative "pre-salt" petroleum system. Additional source rocks (e.g. Toarcian and Kimmeridgian clays) may be present but have not yet been identified. Potential sandstone reservoirs occur in the Middle and Upper Jurassic section. An efficient seal is provided by Upper Jurassic evaporites. Trapping mechanisms are still poorly understood , but the present-day trap geometries are probably relatively young and are related to the Pliocene to Quaternary uplift of the Greater Caucasus (Avdeev and Niemi, 2011).
The high maturity of Jurassic rocks near extrusive sills suggests that a first hydrocarbon generation phase was related to magmatic heating, but these   early hydrocarbons have probably been lost. A second generation phase is probably related to deep burial of source rocks beneath the Upper Miocene and Pliocene succession. Apart from the above oil system, a conventional gas and a coal-bed-methane system may be associated with Bathonian coals mined in the Tkibuli-Shaori region (Fig 1b). According to GIG (2016), the gas content of the high volatile bituminous coal is 15 m³/tons. Moreover, borehole K-1 produced more than 30 million m³ of thermogenic gas from upper Bajocian sediments.

"Post-salt" petroleum system in the Rioni Basin
The "post-salt" petroleum system is based on prolific Kuma and Maikop source rocks, but additional source rocks are present at Lower Cretaceous (e.g. Albian) and lower Paleogene levels. Their relative contributions to the accumulated oils are unknown, but they may have contributed significantly to the Chaladidi oils, which may form a separate oil family in the Rioni Basin. Fractured carbonate and sandstone reservoirs occur in the Upper Cretaceous and Mio-Pliocene successions, respectively. Trap formation in the Achara-Trialeti thrust-fold belt was very young (latest Pliocene to Recent) and slightly older in the northern part of the basin (Maeotian to Pontian). Hydrocarbon generation accompanied deep, Late Miocene to Recent burial of the source rocks. Cumulative production (<200,000 tons) and proven oil reserves (~2 million tons) are low and may be a result of low source rock maturity.

Kura Basin petroleum system
Oil-source correlation in the Kura Basin has only partly been successful. The most likely source rocks are the shales within the very thick Maikop Group, despite its relatively poor (at best "fair") petroleum potential. Yükler et al. (2019) speculated that the source rock quality of the Maikop Group in the Tbilisi area may increase toward the north, where these authors predicted oil and gas generation may occur at depths greater than 3600 and 4350 m, respectively. The kitchen area is probably located to the north of the Norio field and the Manavi discovery (Fig. 1b) contributed to the accumulated hydrocarbons (mainly oil, minor gas). Migration pathways were provided predominantly by strike-slip faults, and migration may have been stimulated by overpressure (Yükler et al., 2019). Trap formation is young and is related to deformation along the front of the Achara-Trialeti and Greater Caucasus fold-thrust belts. In the Tbilisi area, as a result of the very great thicknesses of the Oligocene and Miocene succession, hydrocarbon generation may already have begun during the Early Miocene (e.g. Patton, 1993;Pupp et al., 2018;Corrado et al., 2021). In eastern Georgia, hydrocarbon generation may have started later, probably in the late Pliocene. Nearly 90 % of the cumulative production of the Kura Basin in Georgia (28.5 million tons) is from a single field (Samgori-Patardzeuli-Ninotsminda). Moreover, proven reserves (as of 2014) are only 2.4 million tons.
Considering the high number of oil seeps, this may reflect the negative impact of inversion tectonics on trap integrity.

CONCLUSIONS
Hydrocarbon accumulations in Georgia occur in the western Rioni Basin and the eastern Kura Basin. Both "pre-salt" and "post-salt" petroleum systems can be distinguished in the Rioni Basin. Source rocks of the "pre-salt" petroleum system are poorly defined but it is evident that Bathonian shales have generated liquid hydrocarbons. These are present in Bathonian sandstones (Racha Shaori coal mine), and additional reservoir units include Upper Jurassic sandstones (e.g. Okumi oil discovery) sealed by evaporitic rocks. The origin of the oil at Okumi in the northernmost part of the Rioni Basin is still unknown. The CBM potential of Bathonian coals deserves further consideration.
The "post-salt" petroleum system in the Rioni Basin is based on the Middle Eocene Kuma Formation and the Oligocene part of the Maikop Group. Additional Lower Cretaceous and Paleogene source rocks are present and may have contributed to accumulated hydrocarbons in the Supsa/Shromisubani and Chaladidi fields. The oil is reservoired in Upper Cretaceous fractured carbonates (Chaladidi) or Sarmatian (Shromisubani) and Maeotian sandstones (Supsa). Oil in the shallow Supsa structure reaches level 4 of the biodegradation scale of Peters and Moldowan (1993).